This assignment is for ETC5521 Assignment 1 by Team brolga comprising of Dilinie Seimon and Diyao Chen.

1 Introduction and motivation

Animal Crossing: New Horizons is a life simulation video game developed and published by Nintendo for the Nintendo Switch. It was released worldwide on the 20th of March 2020. Since it’s inception the game has had an astounding world-wide reception With over 22 million copies of the game being sold in just 5 months.

In the animal crossing world a player takes the role of a customized human character who moves to a deserted island and carries out various activities such as gathering and crafting items, fishing and bug hunting in a village inhabited by various species of animals. Each of these animals called villagers in the animal crossing world, have their own name, gender, birthday, personality, favorite song and their own catchphrase. The items used in performing different tasks in the animal crossing belong to different categories and are also priced at different buying and selling values.

The game also simulates day and night based on a 24 hour clock, and has different animals and insects appearing at specific times of day or night. The level of detail in the design of the Animal Crossing world has attracted many users playing throughout the day.

Although it has been 5 only months since the game was released worldwide, it has been gaining a lot of attention from both players and critics. Even non-players may find the concept of this game well designed and interesting.

The motivation for choosing Animal Crossing for the analysis is to attempt to understand the reason for the immense popularity of the game using user and critic reviews and features within the game (such as the villagers and the items).

Therefore, the analysis of the Animal Crossing: New Horizons game is subdivided into two broad areas and answers the following sub-questions.

An analysis of the villagers and items used in the Animal Crossing world

What are players and critics saying about Animal Crossing?

2 Data description

The dataset used for this analysis was retrieved from TidyTuesday; a project aimed at allowing individuals to practice their data wrangling and visualization skills through the use of real-world data sets.

The retrieved dataset consisted of four sub-datasets; ‘villagers’ and ‘items’ datasets containing data about in-game characters and items, and ‘user reviews’ and ‘critic reviews’ datasets containing data about user and critics reviews on the game.

The ‘villagers’ and ‘items’ datasets have been originally retrieved from VillagerDB, which is a project aimed at making data about Animal Crossing available and easily accessible, while the user and critic reviews have been originally retrieved by scraping the Metacritic website.

2.0.1 Data dictionary

Villagers

The ‘Villagers’ dataset consists of data related to the characters in the Animal Crossing game world. The following are the variables in the villagers dataset and their descriptions.

Variable Description
row_n Numeric identifier of villager
id Short text identifier of villager
name Name of villager
gender Gender of villager
species Species of villager
birthday Birthday of villager
personality Personality of Villager
song Song associated with villager
phrase Catchphrase of the villager
full_id Full text identifier of villager
url Link to image of villager

Items

The ‘Items’ dataset consists of data related to the items in the Animal Crossing game world. The following are the variables in the items dataset and their descriptions.

Variable Description
num_id Numeric identifier of item
id Character identifier of villager
name Name of item
category Category of item
orderable Orderable from catalog
sell_value Selling value
sell_currency Selling currency
buy_value Buying value
buy_currency Buying currency
sources Way/place to acquire item
customizable Is item customizable
recipe Recipe of the item - material made of
recipe_id Recipe ID
games_id Game ID
id_full Full Character ID
image_url Link to image of item

User Reviews

The ‘User Reviews’ dataset consists of the scores and reviews made by users from 2020-03-20 to 2020-05-03.

Variable Description
grade Raw score(0-10) given, 0-lowest and 10-highest
user_name User name of the reviewer
text Raw text of the review
date Date the review was published

Critic Reviews

The ‘Critic Reviews’ dataset consists of the scores and reviews made by critics about the game from 2020-03-16 to 2020-05-01.

Variable Description
grade Raw score(0-100) given, 0-lowest and 100-highest
publication The source of the reviewer
text Raw text of the review
date Date the review was published

3 Data Exploration and Wrangling

3.1 Handling Missing Values

It’s interesting how the buying value of 22.21% of the items are missing. In order to analyse this further, the percentages of missing buying values of each item category was calculated.

Figure 3.1: Percentage of missing buying values in each category of items

Figure 3.1 states that all buying values of fish, fossils and seashells are missing.

Further research claimed that fish, fossils and seashells can not be bought, which explains the missing buying values. A blog on Animal Crossing states that fish can only be acquired by fishing and can not be bought (“Animal Crossing: New Horizons Fish Guide: How, When and Where to Catch All the Fish” n.d.). Further, a fan page on Animal Crossing states that fossils can only be dug up and seashells can be collected (“Animal Crossing Wiki” n.d.). The rest of the missing buying values too were attributed to be due to the inability to purchase the items in the Animal Crossing World.

The columns with over 80% missing values were dropped from the analysis due to the inability to impute values accurately. This did not impact the analysis due to the independence of each of the dropped variables from the rest.

3.2 Currency conversion

The buying and selling values of items in the animal crossing world were expressed in two currencies; Bells and Miles. For simplification of the analysis all buying and selling prices were converted into Bells.

The Nintendo guides states that a Bell Voucher can be bought for 500 Nook Miles, which can thereafter be exchanged for 3000 Bells in the Animal Crossing world . Therefore it was assumed that each Mile equated to 6 Miles in the currency conversion.

4 Analysis and findings

4.1 An analysis of the villagers in the Animal Crossing World

Non-player characters (NPC) play an important role in most games, to guide the player through the game and give the player a better virtual experience. In the Animal Crossing world, the non-player characters are called villagers and not only guide the player through the game but also live alongside the player in the game world.

The villagers in Animal Crossing are of different species and also have their own gender, birthdate, personality, favorite song and unique catch phrase. The game is designed such that each villager is an independent individual keeping players attracted to the game for hours.

Figure 4.1 is a plot of the villagers in the Animal Crossing world belonging to each species and gender category.

Figure 4.1: The number of villagers belonging to each species

As visualized in Figure 4.1, there are 391 villagers belonging to 35 different species in the Animal Crossing world. Cats, rabbits, frogs and squirrels are among the most common species, while bulls, rhinos, cows and octopuses are relatively uncommon. Another interesting observation can be seen in the breakdown of each species by gender. Most species have villagers belonging to both male and female categories, but all bulls and lions are male while all cows are female. This maybe due to the terms ‘cow’ and ‘bull’ being gender specific; cows being the female counterpart of bulls. This however does not explain the non-existence of female lions in the Animal Crossing world.

Figure 4.2 is a breakdown of the villagers by their personalities and genders.

Figure 4.2: Count of villagers if each personality type in Animal Crossing

According to figure 4.1, the villagers in the Animal Crossing world belong to either of eight different personality types. It’s interesting how most villagers have normal or lazy personalities, while very few have smug or uchi personalities. The Uchi personality type which is also the rarest personality type, is described as sisterly, tough and caring about their appearance (“Animal Crossing Wiki” n.d.).

Another direct insight from figure 4.1 is that each personality type relates to a single gender. Female villagers in the Animal Crossing World are either normal, peppy, snooty or uchi while males are either cranky, jocky, lazy or smug.

Figure 4.3: The different personallity types of each species

Figure 4.3 is a plot of the number of villagers belonging to each species broken down by their personality types. By the analysis, it is evident that the Animal Crossing world has a diverse set of villagers belonging to different species and genders and also having different personalities.

4.2 An analysis of the Items in the Animal Crossing World

The Animal Crossing world has many different items assisting the villagers in tasks such as building houses, fishing, bug-hunting, digging etc. Each of these items have their own characteristics and belong to one of the 21 categories.

Figure 4.4 visualizes the number of items belonging to each of the item categories. Most of the items in the Animal Crossing are furniture items, while there are also alot of photos. The categories fruit and seashells contain the least number of items, while it’s interesting to see more types of umbrellas and socks than fruits. The Animal Crossing world seems very well equipped with item categories ranging from flooring to umbrellas.

Figure 4.4: Count of items belonging to each category

Figure ?? visualizes the median buying and selling prices of each item category.

It’s interesting how the median selling price of each category is lower than the median buying price except for tools. The median buying prices of fish, fossils and seashells are not available as they are not available for sale in the animal crossing world. The difference between the median buying price and median selling price is proportional to the distance between the two points on each category. Furniture seems to have the highest median profit, while photos seem to have the least.

The most expensive and cheapest items available to be bought in the Animal crossing world are represented in table 4.1 and table 4.2

Table 4.1: Most expensive items available to buy
name category Buying Price Selling Price
Royal Crown Hats 1200000 bells 300000 bells
Crown Hats 1000000 bells 250000 bells
Gold Armor Dresses 320000 bells 80000 bells
Golden Casket Furniture 320000 bells 80000 bells
Grand Piano Furniture 260000 bells 65000 bells
Golden Toilet Furniture 240000 bells 60000 bells
Table 4.2: Cheapest items available to buy
name category Buying Price Selling Price
Admiral’s Photo Photos 40 bells 10 bells
Agent S’s Photo Photos 40 bells 10 bells
Agnes’s Photo Photos 40 bells 10 bells
Alfonso’s Photo Photos 40 bells 10 bells
Alice’s Photo Photos 40 bells 10 bells
Alli’s Photo Photos 40 bells 10 bells

The most expensive items available to buy are crowns and furniture such as armors, caskets, pianos and toilets while the cheapest items are all photos. Upon attempting to analyse the most profitable items based on resale, it was identified that no item in the Animal Crossing world generated a profit on resale.

4.3 What do the users say?

The analysis of user feedback on Animal Crossing uses 2999 reviews published by users on Metacritic from 2020-03-20 - 2020-05-03

Figure 4.5 is a plot of the trend of user reviews on Metacritic over time.

Figure 4.5: Trend of user reviews

The astounding reception of Animal Crossing: New Horizons since its world release on the 20th of March 2020 is justified by the number of daily user reviews it has received. Figure 4.5 shows a huge spike in the number of reviews on the 24th of March 2020, lasting till about the 26th of March 2020, which may be attributed to the world release of the game on the 20th of March 2020. The number of reviews there after remain consistent other than another smaller spike around the 28th of April 2020.

Figure 4.6 shows the most common words in the user reviews for the game. The words ‘game’, ‘island’, ‘switch’ and ‘play’ are the most common words used in the reviews and a direct positive or negative significance can not be obtained based on them.

The most used words in the user reviews

Figure 4.6: The most used words in the user reviews

The user reviews also includes a score from 0-10, where 0 is the lowest and 10 is the highest. Figure 4.7 is a plot of the distribution of scores ranging from 0-10.

Figure 4.7: Distribution of user review scores on Animal Crossing: New Horizons

Most users score the game as a 0, while other users score the game as a 10. Almost all user scores are distributed to the two ends of the range of scores with very little reviews scoring the game a 5, 6 or 7. With the sudden hype about the game in the recent past, the low review scores seem questionable and may even thought of as the default score attached to a review if not explicitly stated.

The top 15 common words in low and high grade reviews

(#fig:high_vs_low, )The top 15 common words in low and high grade reviews

According to the figure @ref(fig:high_vs_low), there is a lot overlapping for the most common words. We can extract some information is, in the high group, the user in the comments putting more games, series, fun, review, love, amazing and 10 and in the low group, the user in the comments putting more buy, experience, person, 1, progress, family and multiple. Only seven words are different. So it’s hard to analyze from common words why users give high marks or low marks. Therefore, it might be interesting to calculate the sentiments of the user reviews and relate them to their respective scores, to identify any correlation among them.

A sentiment score between -5 and +5 are given to each user review, where -5 indicates a highly negative sentiment and +5 indicates a highly positive sentiment.

Table 4.3: Mean sentiments of each grade group
grade sentiment
10 1.1062425
9 1.0766219
8 0.7438356
7 0.7278481
6 0.3278351
5 0.4943343
4 0.2962963
3 0.1961722
2 0.0836177
1 0.0950000
0 -0.0851129

The table 4.3 shows the mean sentiment in the different group grade. We can see very clearly when the higher the mean sentiment is, the higher the user’s rating will be. But for this game the mean sentiment in grade 2 group is higher than grade 1 group probably because the user reviews are low.

Figure 4.8 is a boxplot summarizing the sentiment scores of all 2999 from 2020-03-20 - 2020-05-03.

Summary of sentiments of user-reviews

Figure 4.8: Summary of sentiments of user-reviews

The boxplot in figure 4.8 states that the overall sentiment of the user reviews to be just slightly positive at 0.4, which is surprising as it would’ve been expected to be much higher with the recent popularity it has gotten. Most of the sentiments of the reviews also lie within a range of -1 to +1, which may indicate to us that Animal crossing isn’t enjoyed by all and there are as many users dissatisfied by the game or disliking the game as those enjoying it.

The mean sentiment score of each review against its review score is plotted in figure 4.9.

Figure 4.9: Mean sentiment score of reviews against its review score

The distribution of points over the plot signifies no clear relationship among the sentiments of the review text and score.

4.4 What do the critics say?

In the analysis the viewpoint of critics on the Animal Crossing: New Horizons game, reviews published by 107 critics such as Forbes, Telegraph and Nintendo Life from 2020-03-16 - 2020-05-01 are used.

Figure 4.10: Trend of critic reviews

Figure 4.10 represents the trend in the number of critic reviews over time. Most critics have reviewed the game on the 16th of March, just before the world release of the game, while a smaller number of critics have made reviews in the days following that.

Since critics have the ability to influence people through their comments, it might be interesting to see the most used positive and negative words in their reviews.

The most used positive words by the critics

Figure 4.11: The most used positive words by the critics

The most used negative words by the critics

Figure 4.12: The most used negative words by the critics

Figure 4.11 shows the most used positive words in the critic reviews while figure 4.12 shows the most negative words. By direct observation of the number of terms in the two word clouds, the positivity seems to overpower the negativity in the critics reviews.

Figure 4.13 is a further breakdown of the words used by critics in their reviews, based on different emotions portrayed.

Figure 4.13: Break down of words used by critics into different emotional categories

As per figure 4.13, most words used in critics reviews are positive, and resonate the emotions of trust, anticipation and joy. A very few words used in reviews resonate the emotions of disgust, fear and anger, concluding an overall positive response from critics on the Animal Crossing game.

The scores given by the critics range from 0-100, 0 being the lowest and 100 being the highest. Figure 4.14 shows the distribution of these scores over critics reviews.

Figure 4.14: The distribution of critics scores on Animal Crossing- New Horizon

It’s interesting how a significant percentage of the reviews score the game above 90, while all of the scores are above 70. Comparing figure 4.7 and figure 4.14, all the critics seem impressed with the game while the users have mixed reviews.

5 How do reviews change over time?

We know based off our previous analysis, that Animal Crossing is something of a slow-burn as the game aims to simulate the pace of real life in certain ways. As such, this might affect the ability of reviewers (both critics and users) to fairly assess the game. For instance, how can users conceivably be able to review the game within the week of its release? How can one get a sound grasp of the functionality and the intrinsic rewards of the game without playing it for a few weeks at a minimum?

So, let’s group both user and critic reviews by week and calculate the average review and how it changes as the weeks pass.

Weekly average User and Critic Grades

Figure 5.1: Weekly average User and Critic Grades

Figure 5.1 shows us the Week (measured as the number of weeks since the start of year and the date of measurement) on the x-axis and the weekly average grade given by User of Critic on the y-axis.

The weekly average user grades starts at about 5.5 before beginning a decline in subsequent weeks and hovering at around 3-3.5 through week 16 before a significant jump in week 17 to 5.5 after which it once again drops.

The weekly average critic grades follow a moderately similar shape. The average critic grades are steady in the early 90s for the first couple of weeks before dipping below 90 in week 15. Interestingly, the average grade then experiences a sharp increase in week 17 - similar to what we saw in the weekly average user grades - before declining again.

So, overall the data does not suggest that there is a clear indication that users or critics’ feelings on the game improve or decline as they spend more time with the game. However, there are two things that ought to be explored from this point.

Firstly, we can see that the shape of the both weekly average critic and user grades are roughly similar. But how similar? The correlation coefficent of weekly average user grades and weekly average critic grades is 0.5684725 which indicates a weakish positive relationship between the two. It is not feasible, therefore, to suggest that there might be an influence of critic reviews on user reviews.

The second thing of interest is the dramatic spike seen in critic and user reviews in week 17. Did something happen in week 17 that led to a dramatic shift in the view of the game? Week 17 occured from the the 22nd of April 2020 to the 28th of April 2020. A quick google search of the term ‘Animal Crossing’ with a filter on these dates reveals that Nintendo released an update of the game on April the 23rd. Per Express, the ‘Earth Day’ update included the addition of a myriad of new features which evidently enriched the experience of game.

References

“Animal Crossing Item and Villager Database.” n.d. VillagerDB. Accessed August 26, 2020. https://villagerdb.com/.

“Animal Crossing: New Horizons Fish Guide: How, When and Where to Catch All the Fish.” n.d. GamesRadar+. Accessed August 26, 2020. https://www.gamesradar.com/au/animal-crossing-new-horizons-fish/.

“Animal Crossing: New Horizons Reviews.” n.d. Metacritic. Accessed August 26, 2020. https://www.metacritic.com/game/switch/animal-crossing-new-horizons/critic-reviews.

“Animal Crossing Wiki.” n.d. FANDOM. Accessed August 26, 2020a. https://animalcrossing.fandom.com/wiki/Fossils_(New_Horizons).

“Animal Crossing Wiki.” n.d. FANDOM. Accessed August 26, 2020b. https://animalcrossing.fandom.com/wiki/Sisterly.

Fellows, Ian. 2018. Wordcloud: Word Clouds. https://CRAN.R-project.org/package=wordcloud.

Grolemund, Garrett, and Hadley Wickham. 2011. “Dates and Times Made Easy with lubridate.” Journal of Statistical Software 40 (3): 1–25. http://www.jstatsoft.org/v40/i03/.

Sievert, Carson. 2020. Interactive Web-Based Data Visualization with R, Plotly, and Shiny. Chapman; Hall/CRC. https://plotly-r.com.

Silge, Julia, and David Robinson. 2016. “Tidytext: Text Mining and Analysis Using Tidy Data Principles in R.” JOSS 1 (3). The Open Journal. https://doi.org/10.21105/joss.00037.

“TidyTuesday.” n.d. GamesRadar+. Accessed August 26, 2020. https://github.com/rfordatascience/tidytuesday.

Tierney, Nicholas. 2017. “Visdat: Visualising Whole Data Frames.” JOSS 2 (16). Journal of Open Source Software: 355. https://doi.org/10.21105/joss.00355.

Wickham, Hadley. 2016. Ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York. https://ggplot2.tidyverse.org.

Wickham, Hadley, Mara Averick, Jennifer Bryan, Winston Chang, Lucy D’Agostino McGowan, Romain François, Garrett Grolemund, et al. 2019. “Welcome to the tidyverse.” Journal of Open Source Software 4 (43): 1686. https://doi.org/10.21105/joss.01686.

Wickham, Hadley, Jim Hester, and Romain Francois. 2018. Readr: Read Rectangular Text Data. https://CRAN.R-project.org/package=readr.

Zhu, Hao. 2019. KableExtra: Construct Complex Table with ’Kable’ and Pipe Syntax. https://CRAN.R-project.org/package=kableExtra.